oota-llvm.git
10 years agoDWARFAbbreviationDeclaration: remove dead code, refactor parsing code and make it...
Alexey Samsonov [Thu, 31 Oct 2013 17:20:14 +0000 (17:20 +0000)]
DWARFAbbreviationDeclaration: remove dead code, refactor parsing code and make it more robust. No functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193770 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoLower stackmap intrinsics directly to their target opcode in the DAG builder.
Andrew Trick [Thu, 31 Oct 2013 17:18:24 +0000 (17:18 +0000)]
Lower stackmap intrinsics directly to their target opcode in the DAG builder.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193769 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd Verifier test case for variable argument intrinsics.
Andrew Trick [Thu, 31 Oct 2013 17:18:17 +0000 (17:18 +0000)]
Add Verifier test case for variable argument intrinsics.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193768 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd experimental stackmap intrinsics to definition file and documenation.
Andrew Trick [Thu, 31 Oct 2013 17:18:14 +0000 (17:18 +0000)]
Add experimental stackmap intrinsics to definition file and documenation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193767 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoEnable variable arguments support for intrinsics.
Andrew Trick [Thu, 31 Oct 2013 17:18:11 +0000 (17:18 +0000)]
Enable variable arguments support for intrinsics.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193766 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agowhitespace
Andrew Trick [Thu, 31 Oct 2013 17:18:07 +0000 (17:18 +0000)]
whitespace

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193765 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove another unused flag.
Rafael Espindola [Thu, 31 Oct 2013 15:58:33 +0000 (15:58 +0000)]
Remove another unused flag.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193756 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove unused flag.
Rafael Espindola [Thu, 31 Oct 2013 15:49:39 +0000 (15:49 +0000)]
Remove unused flag.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193752 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRules adjustments in order to build on DragonFly BSD.
Rafael Espindola [Thu, 31 Oct 2013 14:35:00 +0000 (14:35 +0000)]
Rules adjustments in order to build on DragonFly BSD.

Patch by Robin Hahling.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193750 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove the --shrink-wrap option.
Rafael Espindola [Thu, 31 Oct 2013 14:07:59 +0000 (14:07 +0000)]
Remove the --shrink-wrap option.

It had no tests, was unused and was "experimental at best".

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193749 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd AVX512 unmasked integer broadcast intrinsics and support.
Cameron McInally [Thu, 31 Oct 2013 13:56:31 +0000 (13:56 +0000)]
Add AVX512 unmasked integer broadcast intrinsics and support.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193748 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAVX-512: Implemented CMOV for 512-bit vectors
Elena Demikhovsky [Thu, 31 Oct 2013 13:15:32 +0000 (13:15 +0000)]
AVX-512: Implemented CMOV for 512-bit vectors

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193747 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[SystemZ] Automatically detect zEC12 and z196 hosts
Richard Sandiford [Thu, 31 Oct 2013 12:14:17 +0000 (12:14 +0000)]
[SystemZ] Automatically detect zEC12 and z196 hosts

As on other hosts, the CPU identification instruction is priveleged,
so we need to look through /proc/cpuinfo.  I copied the PowerPC way of
handling "generic".

Several tests were implicitly assuming z10 and so failed on z196.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193742 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[AArch64] Make the use of FP instructions optional, but enabled by default.
Amara Emerson [Thu, 31 Oct 2013 09:32:11 +0000 (09:32 +0000)]
[AArch64] Make the use of FP instructions optional, but enabled by default.

This adds a new subtarget feature called FPARMv8 (implied by NEON), and
predicates the support of the FP instructions and registers on this feature.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193739 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agollvm/test/Bitcode/invalid.ll: Tweak expresion to mach "llvm-dis.EXE:"
NAKAMURA Takumi [Thu, 31 Oct 2013 06:21:00 +0000 (06:21 +0000)]
llvm/test/Bitcode/invalid.ll: Tweak expresion to mach "llvm-dis.EXE:"

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193738 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix a use after free on invalid input.
Rafael Espindola [Thu, 31 Oct 2013 04:20:23 +0000 (04:20 +0000)]
Fix a use after free on invalid input.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193737 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix most memory leaks in tablegen.
Rafael Espindola [Thu, 31 Oct 2013 04:07:41 +0000 (04:07 +0000)]
Fix most memory leaks in tablegen.

Found by the valgrind bot.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193736 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoMerge CallGraph and BasicCallGraph.
Rafael Espindola [Thu, 31 Oct 2013 03:03:55 +0000 (03:03 +0000)]
Merge CallGraph and BasicCallGraph.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193734 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoUpdated llvm-cov's OVERVIEW description
Yuchen Wu [Thu, 31 Oct 2013 02:01:24 +0000 (02:01 +0000)]
Updated llvm-cov's OVERVIEW description

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193732 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoLegalize: Improve legalization of long vector extends.
Jim Grosbach [Thu, 31 Oct 2013 00:20:48 +0000 (00:20 +0000)]
Legalize: Improve legalization of long vector extends.

When an extend more than doubles the size of the elements (e.g., a zext
from v16i8 to v16i32), the normal legalization method of splitting the
vectors will run into problems as by the time the destination vector is
legal, the source vector is illegal. The end result is the operation
often becoming scalarized, with the typical horrible performance. For
example, on x86_64, the simple input of:
define void @bar(<16 x i8> %a, <16 x i32>* %p) nounwind {
  %tmp = zext <16 x i8> %a to <16 x i32>
  store <16 x i32> %tmp, <16 x i32>*%p
  ret void
}

Generates:
  .section  __TEXT,__text,regular,pure_instructions
  .section  __TEXT,__const
  .align  5
LCPI0_0:
  .long 255                     ## 0xff
  .long 255                     ## 0xff
  .long 255                     ## 0xff
  .long 255                     ## 0xff
  .long 255                     ## 0xff
  .long 255                     ## 0xff
  .long 255                     ## 0xff
  .long 255                     ## 0xff
  .section  __TEXT,__text,regular,pure_instructions
  .globl  _bar
  .align  4, 0x90
_bar:
  vpunpckhbw  %xmm0, %xmm0, %xmm1
  vpunpckhwd  %xmm0, %xmm1, %xmm2
  vpmovzxwd %xmm1, %xmm1
  vinsertf128 $1, %xmm2, %ymm1, %ymm1
  vmovaps LCPI0_0(%rip), %ymm2
  vandps  %ymm2, %ymm1, %ymm1
  vpmovzxbw %xmm0, %xmm3
  vpunpckhwd  %xmm0, %xmm3, %xmm3
  vpmovzxbd %xmm0, %xmm0
  vinsertf128 $1, %xmm3, %ymm0, %ymm0
  vandps  %ymm2, %ymm0, %ymm0
  vmovaps %ymm0, (%rdi)
  vmovaps %ymm1, 32(%rdi)
  vzeroupper
  ret

So instead we can check if there are legal types that enable us to split
more cleverly when the input vector is already legal such that we don't
turn it into an illegal type. If the extend is such that it's more than
doubling the size of the input we check if
  - the number of vector elements is even,
  - the source type is legal,
  - the type of a split source is illegal,
  - the type of an extended (by doubling element size) source is legal, and
  - the type of that extended source when split is legal.
If the conditions are met, instead of just splitting both the
destination and the source types, we create an extend that only goes up
one "step" (doubling the element width), and the continue legalizing the
rest of the operation normally. The result is that this operates as a
new, more effecient, termination condition for the loop of "split the
operation until the destination type is legal."

With this change, the above example now compiles to:
_bar:
  vpxor %xmm1, %xmm1, %xmm1
  vpunpcklbw  %xmm1, %xmm0, %xmm2
  vpunpckhwd  %xmm1, %xmm2, %xmm3
  vpunpcklwd  %xmm1, %xmm2, %xmm2
  vinsertf128 $1, %xmm3, %ymm2, %ymm2
  vpunpckhbw  %xmm1, %xmm0, %xmm0
  vpunpckhwd  %xmm1, %xmm0, %xmm3
  vpunpcklwd  %xmm1, %xmm0, %xmm0
  vinsertf128 $1, %xmm3, %ymm0, %ymm0
  vmovaps %ymm0, 32(%rdi)
  vmovaps %ymm2, (%rdi)
  vzeroupper
  ret

This generalizes a custom lowering that was added a while back to the
ARM backend. That lowering is no longer necessary, and is removed. The
testcases for it, however, provide excellent ARM tests for this change
and so remain.

rdar://14735100

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193727 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix a few typos
Matt Arsenault [Wed, 30 Oct 2013 23:43:29 +0000 (23:43 +0000)]
Fix a few typos

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193723 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix CodeGen for unaligned loads with address spaces
Matt Arsenault [Wed, 30 Oct 2013 23:30:05 +0000 (23:30 +0000)]
Fix CodeGen for unaligned loads with address spaces

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193721 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoTeach scalarrepl about address spaces
Matt Arsenault [Wed, 30 Oct 2013 22:54:58 +0000 (22:54 +0000)]
Teach scalarrepl about address spaces

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193720 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd calls to doInitialization() and doFinalization() in verifyFunction()
Rafael Espindola [Wed, 30 Oct 2013 22:37:51 +0000 (22:37 +0000)]
Add calls to doInitialization() and doFinalization() in verifyFunction()

The function verifyFunction() in lib/IR/Verifier.cpp misses some
calls. It creates a temporary FunctionPassManager that will run a
single Verifier pass. Unfortunately, FunctionPassManager is no
PassManager and does not call doInitialization() and doFinalization()
by itself. Verifier does important tasks in doInitialization() such as
collecting type information used to check DebugInfo metadata and
doFinalization() does some additional checks. Therefore these checks
were missed and debug info couldn't be verified at all, it just
crashed if the function had some.

verifyFunction() is currently not used in llvm unless -debug option is
enabled, and in unittests/IR/VerifierTest.cpp

VerifierTest had to be changed to create the function in a module from
which the type debug info can be collected.

Patch by Michael Kruse.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193719 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoProduce .weak_def_can_be_hidden for some linkonce_odr values
Rafael Espindola [Wed, 30 Oct 2013 22:08:11 +0000 (22:08 +0000)]
Produce .weak_def_can_be_hidden for some linkonce_odr values

With this patch llvm produces a weak_def_can_be_hidden for linkonce_odr
if they are also unnamed_addr or don't have their address taken.

There is not a lot of documentation about .weak_def_can_be_hidden, but
from the old discussion about linkonce_odr_auto_hide and the name of
the directive this looks correct: these symbols can be hidden.

Testing this with the ld64 in Xcode 5 linking clang reduces the number of
exported symbols from 21053 to 19049.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193718 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDebugInfo: Push header handling down into CompileUnit
David Blaikie [Wed, 30 Oct 2013 20:42:41 +0000 (20:42 +0000)]
DebugInfo: Push header handling down into CompileUnit

This is a preliminary step to handling type units by abstracting over
all (type or compile) units.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193714 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Mips] Add more SHF_MIPS_xxx ELF section flags.
Simon Atanasyan [Wed, 30 Oct 2013 20:41:45 +0000 (20:41 +0000)]
[Mips] Add more SHF_MIPS_xxx ELF section flags.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193713 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd DebugInfo testcase for high_pc encoded as constant, fixed in r193555.
Will Dietz [Wed, 30 Oct 2013 20:27:17 +0000 (20:27 +0000)]
Add DebugInfo testcase for high_pc encoded as constant, fixed in r193555.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193711 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix GVN creating bitcast between address spaces
Matt Arsenault [Wed, 30 Oct 2013 19:05:41 +0000 (19:05 +0000)]
Fix GVN creating bitcast between address spaces

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193710 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoThis commit adds some (but not all) of the x86-64 relocations that are not
Tom Roeder [Wed, 30 Oct 2013 18:47:25 +0000 (18:47 +0000)]
This commit adds some (but not all) of the x86-64 relocations that are not
currently supported in the ELF object writer, along with a simple test case.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193709 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd {start,end}with_lower methods to StringRef.
Rui Ueyama [Wed, 30 Oct 2013 18:32:26 +0000 (18:32 +0000)]
Add {start,end}with_lower methods to StringRef.

startswith_lower is ocassionally useful and I think worth adding.
endwith_lower is added for completeness.

Differential Revision: http://llvm-reviews.chandlerc.com/D2041

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193706 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[ARM] NEON instructions were erroneously decoded from certain invalid encodings
Artyom Skrobov [Wed, 30 Oct 2013 18:10:09 +0000 (18:10 +0000)]
[ARM] NEON instructions were erroneously decoded from certain invalid encodings

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193705 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Custom lower f32 = uint_to_fp i64
Tom Stellard [Wed, 30 Oct 2013 17:22:05 +0000 (17:22 +0000)]
R600: Custom lower f32 = uint_to_fp i64

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193701 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDwarfDebug: Change Abbreviations member from pointer to reference
David Blaikie [Wed, 30 Oct 2013 17:14:24 +0000 (17:14 +0000)]
DwarfDebug: Change Abbreviations member from pointer to reference

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193699 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agofix RST reference in Writing an LLVM Pass
Benjamin Kramer [Wed, 30 Oct 2013 17:09:32 +0000 (17:09 +0000)]
fix RST reference in Writing an LLVM Pass

Currently, instead of showing up as link, it is rendered as

  ...of FunctionPass <writing-an-llvm-pass-FunctionPass>. The...

PR17733. Patch by Tay Ray Chuan!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193698 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd #include of raw_ostream.h to MipsSEISelLowering.cpp
Hans Wennborg [Wed, 30 Oct 2013 16:10:10 +0000 (16:10 +0000)]
Add #include of raw_ostream.h to MipsSEISelLowering.cpp

Fixing this Windows build error:

..\lib\Target\Mips\MipsSEISelLowering.cpp(997) : error C2027: use of undefined type 'llvm::raw_ostream'

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193696 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][msa] Correct definition of bins[lr] and CHECK-DAG-ize related tests
Daniel Sanders [Wed, 30 Oct 2013 15:45:42 +0000 (15:45 +0000)]
[mips][msa] Correct definition of bins[lr] and CHECK-DAG-ize related tests

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193695 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agomake ConstantRange::signExtend() optimal
Nuno Lopes [Wed, 30 Oct 2013 15:36:50 +0000 (15:36 +0000)]
make ConstantRange::signExtend() optimal
the case [x, INT_MIN) was not handled optimally

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193694 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][msa] Added support for matching bmnz, bmnzi, bmz, and bmzi from normal IR...
Daniel Sanders [Wed, 30 Oct 2013 15:20:38 +0000 (15:20 +0000)]
[mips][msa] Added support for matching bmnz, bmnzi, bmz, and bmzi from normal IR (i.e. not intrinsics)

Also corrected the definition of the intrinsics for these instructions (the
result register is also the first operand), and added intrinsics for bsel and
bseli to clang (they already existed in the backend).

These four operations are mostly equivalent to bsel, and bseli (the difference
is which operand is tied to the result). As a result some of the tests changed
as described below.

bitwise.ll:
- bsel.v test adapted so that the mask is unknown at compile-time. This stops
  it emitting bmnzi.b instead of the intended bsel.v.
- The bseli.b test now tests the right thing. Namely the case when one of the
  values is an uimm8, rather than when the condition is a uimm8 (which is
  covered by bmnzi.b)

compare.ll:
- bsel.v tests now (correctly) emits bmnz.v instead of bsel.v because this
  is the same operation (see MSA.txt).

i8.ll
- CHECK-DAG-ized test.
- bmzi.b test now (correctly) emits equivalent bmnzi.b with swapped operands
  because this is the same operation (see MSA.txt).
- bseli.b still emits bseli.b though because the immediate makes it
  distinguishable from bmnzi.b.

vec.ll:
- CHECK-DAG-ized test.
- bmz.v tests now (correctly) emits bmnz.v with swapped operands (see
  MSA.txt).
- bsel.v tests now (correctly) emits bmnz.v with swapped operands (see
  MSA.txt).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193693 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[AArch64] Add support for NEON scalar floating-point compare instructions.
Chad Rosier [Wed, 30 Oct 2013 15:19:37 +0000 (15:19 +0000)]
[AArch64] Add support for NEON scalar floating-point compare instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193691 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRefactor the AVX512 intrinsics. Cluster the intrinsics into the appropriate vector...
Cameron McInally [Wed, 30 Oct 2013 15:19:10 +0000 (15:19 +0000)]
Refactor the AVX512 intrinsics. Cluster the intrinsics into the appropriate vector extension class within the .td file.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193690 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRehash but don't grow when full of tombstones.
Howard Hinnant [Wed, 30 Oct 2013 15:10:54 +0000 (15:10 +0000)]
Rehash but don't grow when full of tombstones.

This problem was found and fixed by José Fonseca in March 2011 for
SmallPtrSet, committed r128566.  But as far as I can tell, all other
llvm hash tables retain the same problem:  the bucket count can grow
without bound while size() remains near constant by repeated
insert/erase cycles that tend to fill the container with tombstones.
Here is a demo that has been reduced to a trivial case:

int
main()
{
   llvm::DenseSet<unsigned> d;
   for (unsigned i = 0; i < 0xFFFFFFF; ++i)
   {
       d.insert(i);
       d.erase(i);
   }
}

While the container size() never grows above 1, the bucket count grows
like this:

nb = 64
nb = 128
nb = 256
nb = 512
nb = 1024
nb = 2048
nb = 4096
nb = 8192
nb = 16384
nb = 32768
nb = 65536
nb = 131072
nb = 262144
nb = 524288
nb = 1048576
nb = 2097152
nb = 4194304
nb = 8388608
nb = 16777216
nb = 33554432
nb = 67108864
nb = 134217728
nb = 268435456

The above program currently consumes a few GB ram.  This patch brings
the memory consumption down by several orders of magnitude, and keeps
the bucket count at 64 for the above test.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193689 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][msa] Added support for matching bins[lr]i.[bhwd] from normal IR (i.e. not...
Daniel Sanders [Wed, 30 Oct 2013 14:45:14 +0000 (14:45 +0000)]
[mips][msa] Added support for matching bins[lr]i.[bhwd] from normal IR (i.e. not intrinsics)

This required correcting the definition of the bins[lr]i intrinsics because
the result is also the first operand.

It also required removing the (arbitrary) check for 32-bit immediates in
MipsSEDAGToDAGISel::selectVSplat().

Currently using binsli.d with 2 bits set in the mask doesn't select binsli.d
because the constant is legalized into a ConstantPool. Similar things can
happen with binsri.d with more than 10 bits set in the mask. The resulting
code when this happens is correct but not optimal.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193687 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][msa] Combine binsri-like DAG of AND and OR into equivalent VSELECT
Daniel Sanders [Wed, 30 Oct 2013 13:51:01 +0000 (13:51 +0000)]
[mips][msa] Combine binsri-like DAG of AND and OR into equivalent VSELECT

(or (and $a, $mask), (and $b, $inverse_mask)) => (vselect $mask, $a, $b).
where $mask is a constant splat. This allows bitwise operations to make use
of bsel.

It's also a stepping stone towards matching bins[lr], and bins[lr]i from
normal IR.

Two sets of similar tests have been added in this commit. The bsel_* functions
test the case where binsri cannot be used. The binsr_*_i functions will
start to use the binsri instruction in the next commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193682 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips] MipsSETargetLowering now reports DAGCombiner changes when using -debug-only...
Daniel Sanders [Wed, 30 Oct 2013 13:31:27 +0000 (13:31 +0000)]
[mips] MipsSETargetLowering now reports DAGCombiner changes when using -debug-only=mips-isel

No test since -debug output is intended for developers and not end-users.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193681 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][msa] Added support for matching splat.[bhw] from normal IR (i.e. not intrinsics)
Daniel Sanders [Wed, 30 Oct 2013 13:07:44 +0000 (13:07 +0000)]
[mips][msa] Added support for matching splat.[bhw] from normal IR (i.e. not intrinsics)

splat.d is implemented but this subtest is currently disabled. This is because
it is difficult to match the appropriate IR on MIPS32. There is a patch under
review that should help with this so I hope to enable the subtest soon.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193680 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRevert "SelectionDAG: Teach the legalizer to split SETCC if VSELECT needs splitting...
Juergen Ributzka [Wed, 30 Oct 2013 06:36:19 +0000 (06:36 +0000)]
Revert "SelectionDAG: Teach the legalizer to split SETCC if VSELECT needs splitting too."

Now Hexagon and SystemZ are not happy with it :-(

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193677 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSelectionDAG: Teach the legalizer to split SETCC if VSELECT needs splitting too.
Juergen Ributzka [Wed, 30 Oct 2013 05:48:18 +0000 (05:48 +0000)]
SelectionDAG: Teach the legalizer to split SETCC if VSELECT needs splitting too.

The Type Legalizer recognizes that VSELECT needs to be split, because the type
is to wide for the given target. The same does not always apply to SETCC,
because less space is required to encode the result of a comparison. As a result
VSELECT is split and SETCC is unrolled into scalar comparisons.

This commit fixes the issue by checking for VSELECT-SETCC patterns in the DAG
Combiner. If a matching pattern is found, then the result mask of SETCC is
promoted to the expected vector mask type for the given target. This mask has
usually the same size as the VSELECT return type (except for Intel KNL). Now the
type legalizer will split both VSELECT and SETCC.

This allows the following X86 DAG Combine code to sucessfully detect the MIN/MAX
pattern. This fixes PR16695, PR17002, and <rdar://problem/14594431>.

Reviewed by Nadav

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193676 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoReformat Makefile. No other changes.
Bill Wendling [Wed, 30 Oct 2013 04:03:03 +0000 (04:03 +0000)]
Reformat Makefile. No other changes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193675 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips] Compute stack alignment on the fly.
Akira Hatanaka [Wed, 30 Oct 2013 02:29:43 +0000 (02:29 +0000)]
[mips] Compute stack alignment on the fly.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193673 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoReformat code with clang-format.
Josh Magee [Wed, 30 Oct 2013 02:25:14 +0000 (02:25 +0000)]
Reformat code with clang-format.

Differential Revision: http://llvm-reviews.chandlerc.com/D2057

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193672 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoStackProtector.h: Fix trailing comments for doxygen. [-Wdocumentation]
NAKAMURA Takumi [Wed, 30 Oct 2013 00:49:39 +0000 (00:49 +0000)]
StackProtector.h: Fix trailing comments for doxygen. [-Wdocumentation]

  s!//<!///<!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193669 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoTrailing whitespace in a comment line.
NAKAMURA Takumi [Wed, 30 Oct 2013 00:49:33 +0000 (00:49 +0000)]
Trailing whitespace in a comment line.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193668 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDebug Info: code clean up.
Manman Ren [Tue, 29 Oct 2013 23:14:15 +0000 (23:14 +0000)]
Debug Info: code clean up.

Use EmitLabelOffsetDifference for handling on darwin platform when
non-darwin platforms use EmitLabelPlusOffset.

Also fix a bug in EmitLabelOffsetDifference where the size is hard-coded
to 4 even though Size is passed in as an argument.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193660 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDebug Info: support for DW_FORM_ref_addr.
Manman Ren [Tue, 29 Oct 2013 22:57:10 +0000 (22:57 +0000)]
Debug Info: support for DW_FORM_ref_addr.

To support ref_addr, we calculate the section offset of a DIE (i.e. offset
of a DIE from beginning of the debug info section). The Offset field in DIE
is currently CU-relative. To calculate the section offset, we add a
DebugInfoOffset field in CompileUnit to store the offset of a CU from beginning
of the debug info section. We set the value in DwarfUnits::computeSizeAndOffset
for each CompileUnit.

A helper function DIE::getCompileUnit is added to return the CU DIE that
the input DIE belongs to. We also add a map CUDieMap in DwarfDebug to help
finding the CU for a given CU DIE.

For a cross-referenced DIE, we first find the CU DIE it belongs to with
getCompileUnit, then we use CUDieMap to get the corresponding CU for the CU DIE.
Adding the section offset of the CU with the CU-relative offset of a DIE gives
us the seciton offset of the DIE.

We correctly emit ref_addr with relocation using EmitLabelPlusOffset when
doesDwarfUseRelocationsAcrossSections is true.

This commit handles the emission of DW_FORM_ref_addr when we have an attribute
with FORM_ref_addr. A follow-on patch will start using ref_addr when adding a
DIEEntry. This commit will be tested and verified in the follow-on patch.

Reviewed off-list by Eric, Thanks.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193658 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDebug Info: instead of calling addToContextOwner which constructs the context
Manman Ren [Tue, 29 Oct 2013 22:49:29 +0000 (22:49 +0000)]
Debug Info: instead of calling addToContextOwner which constructs the context
after the DIE creation, we construct the context first.

Ensure that we create the context before we create a type so that we can add
the newly created type to the parent. Remove last use of addToContextOwner
now that it's not needed.

We use createAndAddDIE to wrap around "new DIE(". Now all shareable DIEs
should be added to their parents right after the creation.

Reviewed off-list by Eric, Thanks.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193657 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoStruct byval cleanup: add helper functions to reduce code duplication.
Manman Ren [Tue, 29 Oct 2013 22:27:32 +0000 (22:27 +0000)]
Struct byval cleanup: add helper functions to reduce code duplication.

Helper functions are added:
emitPostLd: emit a post-increment load operation with given size.
emitPostSt: emit a post-increment store operation with given size.

No functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193656 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[stackprotector] Update the StackProtector pass to perform datalayout analysis.
Josh Magee [Tue, 29 Oct 2013 21:16:16 +0000 (21:16 +0000)]
[stackprotector] Update the StackProtector pass to perform datalayout analysis.

This modifies the pass to classify every SSP-triggering AllocaInst according to
an SSPLayoutKind (LargeArray, SmallArray, AddrOf).  This analysis is collected
by the pass and made available for use, but no other pass uses it yet.

The next patch will make use of this analysis in PEI and StackSlot
passes.  The end goal is to support ssp-strong stack layout rules.

WIP.

Differential Revision: http://llvm-reviews.chandlerc.com/D1789

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193653 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoUpdate comment
Matt Arsenault [Tue, 29 Oct 2013 21:04:19 +0000 (21:04 +0000)]
Update comment

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193651 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoWorkaround MSVC 32-bit miscompile of getCondCodeAction.
Matt Arsenault [Tue, 29 Oct 2013 20:59:29 +0000 (20:59 +0000)]
Workaround MSVC 32-bit miscompile of getCondCodeAction.

Use 32-bit types for the array instead of 64. This should
generally be better anyway.

In optimized + assert builds, I saw a failure when a
cond code / type combination that is never set was loading
a non-zero value and hitting the != Promote assert.

It turns out when loading the 64-bit value to do the shift,
the assembly loads the 2 32-bit halves from non-consecutive
addresses. The address the second half of the loaded uint64_t
doesn't include the offset of the array in the struct. Instead
of being offset + 4, it's just + 4.

I'm not entirely sure why this wasn't observed before.
setCondCodeAction isn't heavily used by the in-tree targets,
and not with the higher valued vector SimpleValueTypes. Only
PPC is using one of the > 32 valued types, and that is probably
never used by anyone on a 32-bit MSVC compiled host.

I ran into this when upgrading LLVM versions, so I guess the
value loaded from the nonsense address happened to work out
before.

No test since I'm not really sure if / how it can be reproduced
with the current in tree targets, and it's not supposed to change
anything.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193650 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemoving a switch statement that contains only a default label. This resolves an...
Aaron Ballman [Tue, 29 Oct 2013 20:40:52 +0000 (20:40 +0000)]
Removing a switch statement that contains only a default label.  This resolves an MSVC warning.  No functional change intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193649 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips] Align the stack to 16-bytes for mfp64.
Akira Hatanaka [Tue, 29 Oct 2013 19:29:03 +0000 (19:29 +0000)]
[mips] Align the stack to 16-bytes for mfp64.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193641 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove declared but not implemented function.
Rafael Espindola [Tue, 29 Oct 2013 18:31:14 +0000 (18:31 +0000)]
Remove declared but not implemented function.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193637 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix common typos in the docs.
Benjamin Kramer [Tue, 29 Oct 2013 17:53:27 +0000 (17:53 +0000)]
Fix common typos in the docs.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193632 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoMove getSymbol to TargetLoweringObjectFile.
Rafael Espindola [Tue, 29 Oct 2013 17:28:26 +0000 (17:28 +0000)]
Move getSymbol to TargetLoweringObjectFile.

This allows constructing a Mangler with just a TargetMachine.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193630 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDebug Info: clean up testing case.
Manman Ren [Tue, 29 Oct 2013 17:27:14 +0000 (17:27 +0000)]
Debug Info: clean up testing case.

Add a tag before the name attribute for readability. Use CHECK-NEXT
instead of CHECK-NOT followed by a CHECK. Add new lines to separate checking
of different DIEs.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193629 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd a helper getSymbol to AsmPrinter.
Rafael Espindola [Tue, 29 Oct 2013 17:07:16 +0000 (17:07 +0000)]
Add a helper getSymbol to AsmPrinter.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193627 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoadd test cases for frameaddr and returnaddr for aarch64
Weiming Zhao [Tue, 29 Oct 2013 17:01:29 +0000 (17:01 +0000)]
add test cases for frameaddr and returnaddr for aarch64

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193626 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[AArch64] Implement FrameAddr and ReturnAddr
Weiming Zhao [Tue, 29 Oct 2013 17:00:25 +0000 (17:00 +0000)]
[AArch64] Implement FrameAddr and ReturnAddr

Fixes PR17690

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193625 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[ARM] Make sure HasCRC is initialized to false in Subtarget.
Amara Emerson [Tue, 29 Oct 2013 16:54:52 +0000 (16:54 +0000)]
[ARM] Make sure HasCRC is initialized to false in Subtarget.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193624 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSupport for microMIPS jump instructions
Zoran Jovanovic [Tue, 29 Oct 2013 16:38:59 +0000 (16:38 +0000)]
Support for microMIPS jump instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193623 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Add compute support for CI v2
Tom Stellard [Tue, 29 Oct 2013 16:37:28 +0000 (16:37 +0000)]
R600/SI: Add compute support for CI v2

v2:
  - Fix LDS size calculation

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193621 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Expand vector FSQRT ops
Tom Stellard [Tue, 29 Oct 2013 16:37:20 +0000 (16:37 +0000)]
R600: Expand vector FSQRT ops

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193620 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDWARF parser: propery handle DW_FORM_ref_sig8 and fix Windows build.
Alexey Samsonov [Tue, 29 Oct 2013 16:32:19 +0000 (16:32 +0000)]
DWARF parser: propery handle DW_FORM_ref_sig8 and fix Windows build.

Based on D2050 by Timur Iskhodzhanov.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193619 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoThe asm printer has a mangler. Use it.
Rafael Espindola [Tue, 29 Oct 2013 16:24:21 +0000 (16:24 +0000)]
The asm printer has a mangler. Use it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193618 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoThe AsmPrinter has a Mangler. Use it.
Rafael Espindola [Tue, 29 Oct 2013 16:18:15 +0000 (16:18 +0000)]
The AsmPrinter has a Mangler. Use it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193617 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoThe asm printer has a mangler. Don't keep a second pointer to it.
Rafael Espindola [Tue, 29 Oct 2013 16:11:22 +0000 (16:11 +0000)]
The asm printer has a mangler. Don't keep a second pointer to it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193616 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSupport names like llvm-ar-3.4 and llvm-ranlib-3.4.
Rafael Espindola [Tue, 29 Oct 2013 14:25:43 +0000 (14:25 +0000)]
Support names like llvm-ar-3.4 and llvm-ranlib-3.4.

They are used in some packages. For example:
http://packages.ubuntu.com/saucy/i386/llvm-3.4/filelist

This fixes pr17721.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193612 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoTest cleanup for v8 instructions
Bernard Ogden [Tue, 29 Oct 2013 14:16:09 +0000 (14:16 +0000)]
Test cleanup for v8 instructions

Add some missing tests, factor out a test not specific to v8 into
its own file.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193611 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoClarify that GlobalVariables definitions must have an initializer.
Rafael Espindola [Tue, 29 Oct 2013 13:44:11 +0000 (13:44 +0000)]
Clarify that GlobalVariables definitions must have an initializer.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193609 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoQuick-fix DebugInfo build on Windows
Timur Iskhodzhanov [Tue, 29 Oct 2013 12:13:22 +0000 (12:13 +0000)]
Quick-fix DebugInfo build on Windows

MSVC can't comprehend
  template<typename T, size_t N>
  ArrayRef<T> makeArrayRef(const T (&Arr)[N]) {
    return ArrayRef<T>(Arr);
  }
if Arr is
  static const uint8_t sizes[];
declared in a templated and defined a few lines later.

I'll send a proper fix (i.e. get rid of unnecessary templates) for review soon.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193604 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoARM: Add subtarget feature for CRC
Bernard Ogden [Tue, 29 Oct 2013 09:47:35 +0000 (09:47 +0000)]
ARM: Add subtarget feature for CRC

Adds a subtarget feature for the CRC instructions (optional in v8-A) to the ARM (32-bit) backend.

Differential Revision: http://llvm-reviews.chandlerc.com/D2036

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193599 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix misapplied patch in r193597
Anders Waldenborg [Tue, 29 Oct 2013 09:37:28 +0000 (09:37 +0000)]
Fix misapplied patch in r193597

Sorry Peter Zotov, entirely my fault.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193598 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agollvm-c: Make LLVM{Get,Set}Alignment work on {Load,Store}Inst too
Anders Waldenborg [Tue, 29 Oct 2013 09:02:02 +0000 (09:02 +0000)]
llvm-c: Make LLVM{Get,Set}Alignment work on {Load,Store}Inst too

Patch by Peter Zotov

Differential Revision: http://llvm-reviews.chandlerc.com/D1910

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193597 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAArch64: add 'a' inline asm operand modifier
Tim Northover [Tue, 29 Oct 2013 08:22:33 +0000 (08:22 +0000)]
AArch64: add 'a' inline asm operand modifier

This is used in the Linux kernel, and effectively just means "print an
address".

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193593 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDebug Info: instead of calling addToContextOwner which constructs the context
Manman Ren [Tue, 29 Oct 2013 05:49:41 +0000 (05:49 +0000)]
Debug Info: instead of calling addToContextOwner which constructs the context
after the DIE creation, we construct the context first.

This touches creation of namespaces and global variables. The purpose is to
handle all DIE creations similarly: constructs the context first, then creates
the DIE and immediately adds the DIE to its parent.

We use createAndAddDIE to wrap around "new DIE(".

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193589 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd llvm/test/Transforms/SLPVectorizer/ARM/lit.local.cfg. Tests there require ARM...
NAKAMURA Takumi [Tue, 29 Oct 2013 02:46:00 +0000 (02:46 +0000)]
Add llvm/test/Transforms/SLPVectorizer/ARM/lit.local.cfg. Tests there require ARM in targets.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193580 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix "existant" typos
Alp Toker [Tue, 29 Oct 2013 02:35:28 +0000 (02:35 +0000)]
Fix "existant" typos

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193579 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoClean up.
Richard Smith [Tue, 29 Oct 2013 01:44:23 +0000 (01:44 +0000)]
Clean up.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193576 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDWARFFormValue.cpp: Appease gcc to give explicit constructors.
NAKAMURA Takumi [Tue, 29 Oct 2013 01:43:05 +0000 (01:43 +0000)]
DWARFFormValue.cpp: Appease gcc to give explicit constructors.

  error: conversion from `const uint8_t*' to non-scalar type `llvm::ArrayRef<unsigned char>' requested

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193575 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoARM cost model: Unaligned vectorized double stores are expensive
Arnold Schwaighofer [Tue, 29 Oct 2013 01:33:57 +0000 (01:33 +0000)]
ARM cost model: Unaligned vectorized double stores are expensive

Updated a test case that assumed that <2 x double> would vectorize to use
<4 x float>.

radar://15338229

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193574 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoARM cost model: Account for zero cost scalar SROA instructions
Arnold Schwaighofer [Tue, 29 Oct 2013 01:33:53 +0000 (01:33 +0000)]
ARM cost model: Account for zero cost scalar SROA instructions

By vectorizing a series of srl, or, ... instructions we have obfuscated the
intention so much that the backend does not know how to fold this code away.

radar://15336950

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193573 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSLPVectorizer: Use vector type for vectorized memory operations
Arnold Schwaighofer [Tue, 29 Oct 2013 01:33:50 +0000 (01:33 +0000)]
SLPVectorizer: Use vector type for vectorized memory operations

No test case, because with the current cost model we don't see a difference.
An upcoming ARM memory cost model change will expose and test this bug.

radar://15332579

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193572 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoCleaning up comments in lli
Andrew Kaylor [Tue, 29 Oct 2013 01:33:14 +0000 (01:33 +0000)]
Cleaning up comments in lli

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193571 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdding a workaround for __main linking with remote lli and Cygwin/MinGW
Andrew Kaylor [Tue, 29 Oct 2013 01:29:56 +0000 (01:29 +0000)]
Adding a workaround for __main linking with remote lli and Cygwin/MinGW

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193570 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoMove the STT_FILE symbols out of the normal symbol table processing for
Joerg Sonnenberger [Tue, 29 Oct 2013 01:06:17 +0000 (01:06 +0000)]
Move the STT_FILE symbols out of the normal symbol table processing for
ELF. They can overlap with the other symbols, e.g. if a source file
"foo.c" contains a function "foo" with a static variable "c".

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193569 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDebug Info: use createAndAddDIE to wrap around "new DIE" in DwarfDebug.
Manman Ren [Tue, 29 Oct 2013 01:03:01 +0000 (01:03 +0000)]
Debug Info: use createAndAddDIE to wrap around "new DIE" in DwarfDebug.

This commit ensures DIEs are constructed within a compile unit and
immediately added to their parents.

Reviewed off-list by Eric.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193568 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDebug Info: use createAndAddDIE for newly-created Subprogram DIEs.
Manman Ren [Tue, 29 Oct 2013 00:58:04 +0000 (00:58 +0000)]
Debug Info: use createAndAddDIE for newly-created Subprogram DIEs.

More patches will be submitted to convert "new DIE(" to use createAddAndDIE in
DwarfCompileUnit.cpp. This will simplify implementation of addDIEEntry where
we have to decide between ref4 and ref_addr, because DIEs that can be shared
across CU will be added to a CU already.

Reviewed off-list by Eric.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193567 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDebug Info: add a helper function createAndAddDIE.
Manman Ren [Tue, 29 Oct 2013 00:53:03 +0000 (00:53 +0000)]
Debug Info: add a helper function createAndAddDIE.

It wraps around "new DIE(" and handles the bookkeeping part of the newly-created
DIE. It adds the DIE to its parent, and calls insertDIE if necessary. It makes
sure that bookkeeping is done at the earliest time and we should not see
parentless DIEs if all constructions of DIEs go through this helper function.

Later on, we can use an allocator for DIE allocation, and will only need to
change createAndAddDIE instead of modifying all the "new DIE(".

Reviewed off-list by Eric.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193566 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoMerge DWARFDIE::extractFast and DWARFDIE::extract into one function.
Alexey Samsonov [Mon, 28 Oct 2013 23:58:58 +0000 (23:58 +0000)]
Merge DWARFDIE::extractFast and DWARFDIE::extract into one function.

Complicated CU-DIE-specific logic in the latter was never used,
and it makes sense to have safety checks for broken dwarf in the former.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193563 91177308-0d34-0410-b5e6-96231b3b80d8